How el-ouardy/atlaslingua works
Main Components
• Frontend Web Client
– React + TypeScript application (src/)
– UI library: TailwindCSS / Shadcn/ui
– Bundler/dev server: Vite (vite.config.ts)
• Backend API
– Flask application (api/app/)
– REST endpoints organized by feature (routes/*.py)
– Business logic in services/
– Data validation in schemas/
• Datastores
– Relational DB: MySQL (SQLAlchemy models + Alembic migrations)
– Firestore (Firebase) for community forum data
• Serverless Functions
– Firebase Functions (functions/src/index.ts) enforcing Firestore rules
• Third-Party Services
– Algolia search indexing
– SendGrid (email_service.py)
– (Implicit) LLM provider in language_model.py (e.g. OpenAI)
• Hosting & Deployment
– Frontend: Vercel / Netlify / Firebase Hosting
– Backend: Heroku (Procfile) or any container orchestration
– Firebase: managed by firebase.json / .firebaserc
Relationships & Data Flows
• User → Browser → React App
• React App → HTTP(S) → Flask REST API (CRUD on users, history, favorites, contributions, translation/summarization requests)
• Flask API → MySQL (SQLAlchemy) for persistent records (users, dictionaries, history, feedback, etc.)
• Flask API ↔ Firebase Admin SDK (firebase_service.py) for community forum data (posts, comments)
• React App → Firebase SDK → Firestore (community forum UI)
• React App → Algolia API for search queries (dictionary/community)
• Flask API → SendGrid API for transactional emails (password reset, notifications)
• Flask API → LLM Provider (language_model.py) for translation & summarization
Architectural Patterns
• Layered architecture in Backend
– Routing layer (routes/)
– Service layer (services/)
– Data layer (models/, schemas/)
– Utility layer (utils/)
• Separation of concerns in Frontend
– UI components (components/)
– State management (contexts/)
– API clients (services/api.ts)
– Domain models (models/)
• Micro-service-like split between Flask API and Firebase Functions